Experimentation and Characterization of Mobile Broadband Networks

The Internet has brought substantial changes to our life as the main tool to access a large variety of services and applications. Internet distributed nature and technological improvements lead to new challenges for researchers, service providers, and network administrators. Internet traffic measurement and analysis is one of the most trivial and powerful tools to study such a complex environment from different aspects. Mobile BroadBand (MBB) networks have become one of the main means to access the Internet. MBB networks are evolving at a rapid pace with technology enhancements that promise drastic improvements in capacity, connectivity, and coverage, i.e., better performance in general. Open experimentation with operational MBB networks in the wild is currently a fundamental requirement of the research community in its endeavor to address the need for innovative solutions for mobile communications. There is a strong need for objective data relating to stability and performance of MBB (e.g., 2G, 3G, 4G, and soon-to-come 5G) networks and for tools that rigorously and scientifically assess their performance. Thus, measuring end user performance in such an environment is a challenge that calls for large-scale measurements and profound analysis of the collected data. The intertwining of technologies, protocols, and setups makes it even more complicated to design scientifically sound and robust measurement campaigns. In such a complex scenario, the randomness of the wireless access channel coupled with the often unknown operator configurations makes this scenario even more challenging. In this thesis, we introduce the MONROE measurement platform: an open access and flexible hardware-based platform for measurements on operational MBB networks. The MONROE platform enables accurate, realistic, and meaningful assessment of the performance and reliability of MBB networks. We detail the challenges we overcame while building and testing the MONROE testbed and argue our design and implementation choices accordingly. Measurements are designed to stress performance of MBB networks at different network layers by proposing scalable experiments and methodologies. We study: (i) Network layer performance, characterizing and possibly estimating the download speed offered by commercial MBB networks; (ii) End users’ Quality of Experience (QoE), specifically targeting the web performance of HTTP1.1/TLS and HTTP2 on various popular web sites; (iii) Implication of roaming in Europe, understanding the roaming ecosystem in Europe after the "Roam like Home" initiative; and (iv) A novel adaptive scheduler family with deadline is proposed for multihomed devices that only require a very coarse knowledge of the wireless bandwidth. Our results comprise different contributions in the scope of each research topic. To put it in a nutshell, we pinpoint the impact of different network configurations that further complicate the picture and hopefully contribute to the debate about performance assessment in MBB networks. The MBB users web performance shows that HTTP1.1/TLS is very similar to HTTP2 in our large-scale measurements. Furthermore, we observe that roaming is well supported for the monitored operators and the operators using the same approach for routing roaming traffic. The proposed adaptive schedulers for content upload in multihomed devices are evaluated in both numerical simulations and real mobile nodes. Simulation results show that the adaptive solutions can effectively leverage the fundamental tradeoff between the upload cost and completion time, despite unpredictable variations in available bandwidth of wireless interfaces. Experiments in the real mobile nodes provided by the MONROE platform confirm the findings.